This article is more than 5 years old. Information may no longer be current.
Expert: CABG still superior to PCI for left main CAD
Click Here to Manage Email Alerts
ANAHEIM, Calif. — Despite evidence from the EXCEL trial suggesting that PCI is noninferior to CABG in patients with left main CAD, the data are not yet compelling enough to change treatment guidelines for this patient population, a speaker said at the American Heart Association Scientific Sessions.
The EXCEL trial, initially presented at TCT 2016 and published in The New England Journal of Medicine, showed that PCI with an everolimus-eluting stent (Xience, Abbott Vascular) was noninferior to CABG in patients with left main CAD and low or intermediate SYNTAX scores.
Before implementing changes in clinical practice, however, the data require a closer look, according to Marc Ruel, MD, MPH, FRCSC, FAHA, FCCS, M. Pitfield Professor and Chair of the division of cardiac surgery at the University of Ottawa Heart Institute in Canada.
Putting EXCEL in perspective
During his presentation, Ruel highlighted several concerns about the EXCEL trial.
For instance, the study was industry-funded. In such trials, as was the case in EXCEL, CABG is often performed by average centers, whereas PCI is often performed at expert centers, which leads to an inherent site selection bias that can potentially affect the trial’s outcomes, he noted. Moreover, epidemiological data have shown that industry-funded trials are significantly more likely to be positive, with an even higher likelihood for studies with a noninferiority design, according to Ruel.
Another concern involves the patients included in EXCEL, Ruel said. The trial planned to enroll 2,634 patients, but slow recruitment prompted researchers to abandon it prematurely so that the final enrollment included 1,905 patients.
“The question certainly arises as to who are these patients, considering the vast number of patients with left main coronary artery stenoses who are seen by these institutions,” he said.
Additionally, the data have led some to argue that PCI is equivalent to CABG across subgroups, but the subgroup analyses, Ruel said, were considerably underpowered.
In terms of medical therapy, there were also differences between the CABG and PCI study arms, according to Ruel. About 40% of patients had ACS, with one-quarter not even taking aspirin at the time of CABG. At the time, the recommendation for dual antiplatelet therapy for 1 year had not yet been adopted and was more often applied in the PCI group than the CABG group, he said.
EXCEL vs. NOBLE
Physicians and researchers must also consider results from the NOBLE trial, which painted a different picture of PCI vs. CABG for patients with left main CAD. The trial, which was also presented at TCT 2016 and simultaneously published in The Lancet, gave CABG the edge in this patient population.
The differences in these findings, Ruel said, may be attributable to several factors. For instance, one of the major differences involved the definition of the composite primary endpoint.
In EXCEL, Ruel noted, the results demonstrating the noninferiority of PCI when compared with CABG were largely driven by the definition of perioperative MI, which was based on a new definition from the Society for Cardiovascular Angiography and Interventions. The SCAI committee defined “clinically relevant” MI based on enzyme elevation, according to Ruel.
“This new definition was based on small, retrospective, observational studies and certainly the mechanistic inference here is elusive at best,” Ruel said.
The data from EXCEL compared with NOBLE, he noted, showed about half the incidence of perioperative MI in the PCI group vs. the CABG group. This previously had never been hypothesized as a major source of clinical difference between PCI and CABG, but was the major driver of noninferiority in EXCEL, according to Ruel.
He also said the NOBLE trial is often not discussed, but the results are arguably better than those of the EXCEL trial. One potential reason why NOBLE may be left out of the discussion involves its use of a biolimus-eluting stent (BioMatrix Flex, Biosensors) for PCI, whereas an EES was used in EXCEL, according to Ruel.
“However, if you look at the event rate in NOBLE, it’s extremely tight and favorable,” he said. “In fact, outcomes in the trial arms were really amazing.”
Perioperative MI events were also not counted in NOBLE because it was not felt to be a source of important clinical differences, but, in contrast to EXCEL, target vessel revascularization was counted, according to Ruel.
He also noted that EXCEL had a few “firsts,” including the use of the new SCAI perioperative MI definition. Additionally, TVR was not included, which was not the case in previous trials. It was a singular outcome when ischemia-driven revascularization occurred, according to Ruel, therefore implying an adjudication process, which is never fully masked. Another problem was that if TVR is not considered, there should be some correlate of its potential consequences, either with regards to left ventricular function or effects on the native disease progression or restenosis itself, he said.
Further concerns
Most importantly, however, Ruel said more than one trial is necessary to change clinical practice, especially in light of other outcomes from the EXCEL trial.
More recently, data from the EXCEL QOL substudy, presented at TCT 2017 and simultaneously published in the Journal of the American College of Cardiology, showed that quality of life was similar for patients who underwent PCI compared with CABG.
Despite these findings, Ruel said he remains concerned about the hard endpoints. For instance, Ruel said, there was a nonsignificant increase that bordered on a trend toward increased all-cause mortality in the PCI arm. The researchers concluded that the increased mortality was largely attributable to non-CV causes; however, Ruel said adjudication is not a masked process, as it involves looking at the patient journey and, therefore, it is difficult to determine what is and is not CV-related.
“We have seen this before in the FREEDOM trial where the famous P value of .049 was not for the primary outcome but rather one of the secondary outcomes, namely, overall mortality,” he said. “FREEDOM taught us that it’s very hard to argue with death.”
In the same vein, Ruel noted, it is hard to know how EXCEL will play out at 4 and 5 years because it has been decided that that data will not be unmasked for at least another year.
“Before we start talking about equivalent quality of life, as was presented at TCT 2017, we should make sure we have equivalent quantity of life between the two treatment modalities,” Ruel said. “As we know, the landmark analysis, which was mandated by The New England Journal of Medicine, analyzed EXCEL without the definition of perioperative MI. From 30 days to 3 years, the analysis clearly showed that PCI was inferior to CABG for death, MI and stroke with a 44% excess risk and P value of .02.”
In light of these concerns, longer follow-up is essential, according to Ruel.
“We need 5-year data, which we will have, but before then, it would be an absolute mistake to change guidelines based on one noninferiority, industry-funded study and shorter-term follow-up of two studies,” Ruel said. – by Melissa Foster
References:
Ruel M. SA.CVS.2269. Controversies in Coronary Revascularization. Presented at: American Heart Association Scientific Sessions; Nov. 11-15, 2017; Anaheim, Calif.
Baron SJ, et al. J Am Coll Cardiol. 2017;doi:10.1016/j.jacc.2017.10.036.
Mäkikallio T, et al. Lancet. 2016;doi:10.1016/S0140-6736(16)32052-9.
Stone GW, et al. N Engl J Med. 2016;doi:10.1056/NEJMoa1610227.
Disclosure: Ruel reports he has received research grants from Abbott, Edwards Lifesciences, Medtronic and Sanofi-Aventis. The EXCEL trial was supported by Abbott Vascular.